CTCF modulates allele-specific sub-TAD organization and imprinted gene activity at the mouse Dlk1-Dio3 and Igf2-H19 domains.

BACKGROUND: Genomic imprinting is essential for mammalian development and provides a unique paradigm to explore intra-cellular differences in chromatin configuration. So far, the detailed allele-specific chromatin organization of imprinted gene domains has mostly been lacking. Here, we explored the chromatin structure of the two conserved imprinted domains controlled by paternal DNA methylation imprints-the Igf2-H19 and Dlk1-Dio3 domains-and assessed the involvement of the insulator protein CTCF in mouse cells. RESULTS: Both imprinted domains are located within overarching topologically associating domains (TADs) that are similar on both parental chromosomes. At each domain, a single differentially methylated region is bound by CTCF on the maternal chromosome only, in addition to multiple instances of bi-allelic CTCF binding. Combinations of allelic 4C-seq and DNA-FISH revealed that bi-allelic CTCF binding alone, on the paternal chromosome, correlates with a first level of sub-TAD structure. On the maternal chromosome, additional CTCF binding at the differentially methylated region adds a further layer of sub-TAD organization, which essentially hijacks the existing paternal-specific sub-TAD organization. Perturbation of maternal-specific CTCF binding site at the Dlk1-Dio3 locus, using genome editing, results in perturbed sub-TAD organization and bi-allelic Dlk1 activation during differentiation. CONCLUSIONS: Maternal allele-specific CTCF binding at the imprinted Igf2-H19 and the Dlk1-Dio3 domains adds an additional layer of sub-TAD organization, on top of an existing three-dimensional configuration and prior to imprinted activation of protein-coding genes. We speculate that this allele-specific sub-TAD organization provides an instructive or permissive context for imprinted gene activation during development.

Prediction of Drug Approval After Phase I Clinical Trials in Oncology: RESOLVED2.

PURPOSE: Drug development in oncology currently is facing a conjunction of an increasing number of antineoplastic agents (ANAs) candidate for phase I clinical trials (P1CTs) and an important attrition rate for final approval. We aimed to develop a machine learning algorithm (RESOLVED2) to predict drug development outcome, which could support early go/no-go decisions after P1CTs by better selection of drugs suitable for further development. METHODS: PubMed abstracts of P1CTs reporting on ANAs were used together with pharmacologic data from the DrugBank5.0 database to model time to US Food and Drug Administration (FDA) approval (FDA approval-free survival) since the first P1CT publication. The RESOLVED2 model was trained with machine learning methods. Its performance was evaluated on an independent test set with weighted concordance index (IPCW). RESULTS: We identified 462 ANAs from PubMed that matched with DrugBank5.0 (P1CT publication dates 1972 to 2017). Among 1,411 variables, 28 were used by RESOLVED2 to model the FDA approval-free survival, with an IPCW of 0.89 on the independent test set. RESOLVED2 outperformed a model that was based on efficacy/toxicity (IPCW, 0.69). In the test set at 6 years of follow-up, 73% (95% CI, 49% to 86%) of drugs predicted to be approved were approved, whereas 92% (95% CI, 87% to 98%) of drugs predicted to be nonapproved were still not approved (log-rank P < .001). A predicted approved drug was 16 times more likely to be approved than a predicted nonapproved drug (hazard ratio, 16.4; 95% CI, 8.40 to 32.2). CONCLUSION: As soon as P1CT completion, RESOLVED2 can predict accurately the time to FDA approval. We provide the proof of concept that drug development outcome can be predicted by machine learning strategies.